LINE Fukuoka promotes Digital Transformation (DX) within the company through the use of AI, IoT, RPA, and more. As a part of promoting DX, our company developed a workload forecast¹ system that utilizes AI technology.
Beta testing of the system resulted in an 83% reduction in time spent on forecasting work, and we are pleased to announce that the system is planned to be fully released within the Customer Care Department in July.
The same system can also forecast other elements like product demand and recruitment numbers regardless of field, so it's planned to be introduced into many departments in order to increase productivity.
¹Workload forecasting is the prediction of the number of tickets expected to be received in customer care based on past results and recent trends. On the basis of these forecasts, plans for the necessary number of employees and their work schedules are created.
■Key points to the AI-based workload forecast system
・A single forecast system for multiple departments
Many departments forecast workloads for a variety of scenarios, such as recruitment numbers for HR, product orders in sales, and inquiries in customer care. Software used for work improvement is compartmentalized for each task, but this system enables forecasts to be made on a unified platform.
・Reduction in time spent on forecast work
Until now, calculations and analysis were performed manually, and the previous 60 minute average required for forecasting work has been reduced to about ten minutes².
²Required work time may change depending on what's being forecast.
・Elimination of dependence on individual employees for forecast work
Previously, forecasting work required high level analysis skills and experience. Now even employees who have no experience with data analysis can make precise forecasts by using the new system.
・ AI adjustment and learning without the need to rely on an expert
In order to increase the accuracy of forecasts through AI, forecast model adjustments that were previously carried out by experts like data scientists have been automated through this new system. This enables employees who don't have expertise in these fields to adjust AI learning and forecast models on-site.
■Background of the system's introduction
Generally, in customer care personnel is managed based on workload forecasts. Because the amount of inquiries changes often due to seasonal factors such as new services and features being launched or people upgrading to a new mobile device when it's released, accurate workload forecast skills are necessary to preserve the quality of support.
Calculating these forecasts manually based on an enormous amount of past data requires a lot of time and effort. Some aspects of the process are dependent on the "experience and intuition" of the person making the forecast, so becoming dependent on these employees was an issue.
In order to make more precise workload forecasts and create an environment that allows anyone to easily create a forecast regardless of experience, our departments³ that are in charge of internal DX and improving work efficiency began developing this AI-based workload forecast system. Beta testing for the system began in June 2020, resulting in increased forecast accuracy, and by being able to properly calculate expected workloads the average support time for operators stabilized.
³Development of the system was led by our internal Workforce Business Partner Team (a team that creates plans for utilizing and managing our workforce more effectively based on data) and LINE Fukuoka Data Labs (a team that promotes growth and improvement in the operation of LINE services through the use of statistics and machine learning)
■Other examples of DX in the Customer Care Department
LINE Fukuoka's Customer Care Department is responsible for the operation work of many LINE services, and promotes DX in order to offer high quality support without becoming dependent on sheer numbers or overtime.
Representative example: Man-Hour Management System
The "Man-Hour Management System" was built to record and total the man-hours that previously had to be manually recorded on Excel with the click of a button. Business intelligence (BI) tools were utilized to convert registered man-hours into a dashboard, so that operators can manage their productivity in real time.
By introducing this system, the time required to calculate the man-hours from each month was reduced from 30 minutes to just one minute. The Man-Hour Management System isn't just used in Fukuoka, it's used by LINE Group offices all over the world.
In the future, LINE Fukuoka will continue to strive to promote DX within the LINE Group and with companies across Japan by sharing examples of our DX and their results.
Fukuoka will continue to strive to promote DX within the LINE Group and with companies across Japan by sharing examples of our DX and their results.